List your company

Mathematical Reasoning on AIME 2025 (AIME25, Math Avg)

48.38AIME25 ScorePSFT
-1.831211.204424.2437.2756Aug 25, 2025Sep 28, 2025Nov 1, 2025Dec 5, 2025Jan 8, 2026Feb 11, 2026Mar 17, 2026

Evaluation results

MethodLinks
PSFT2025.0848.38—
SFT2025.0846.88—
Granite-3.3 (8B)Variant=PRISM2026.0327.9648.75
Mistral-Small (24B)Variant=PRISM2026.0327.3447.02
Mistral-7BVariant=PRISM2026.0324.2741.28
Granite-4-H Micro (3B)Variant=PRISM2026.0323.4946.64
Nemotron-H-8k (8B)Variant=PRISM2026.0322.7639.53
Granite-4 Micro (3B)Variant=PRISM2026.0322.2942.96
LLaMA-3.1 (8B)Variant=PRISM2026.0319.3236.41
Granite-4 Micro (3B)Variant=Base2026.0312.3426.28
Granite-4-H Micro (3B)Variant=Base2026.032.713.32
Nemotron-H-8k (8B)Variant=Base2026.032.2917.96
Mistral-Small (24B)Variant=Base2026.030.739.48
Granite-3.3 (8B)Variant=Base2026.030.318.95
LLaMA-3.1 (8B)Variant=Base2026.030.152.24
Mistral-7BVariant=Base2026.030.10.59